Cardiovascular Outcomes Comparison of Dipeptidyl Peptidase-4 Inhibitors versus Sulfonylurea as Add-on Therapy for Type 2 Diabetes Mellitus: a Meta-Analysis.

OBJECTIVE: Recent studies have raised concern about the cardiovascular safety of dipeptidyl peptidase-4 (DPP4) inhibitors. We performed a systematic review through meta-analysis to compare cardiovascular outcomes of sulfonylurea (SU) versus DPP4 inhibitors when used in combination with metformin.
METHODS: After searching for trials using combination therapy of metformin with DPP4 inhibitor or SU in PubMed, Cochrane Library, and Embase, one prospective observation study and 15 randomized controlled studies were selected.
RESULTS: Regarding the primary analysis endpoint, there were no significant differences in the risk of all-cause mortality between SU and DPP4 inhibitors as an add-on therapy to metformin (random-effect relative risk [RR], 1.14; 95% confidence interval [CI], 0.98-1.33; p=0.811; I2=0%). Cardiovascular death was also similar between the two drug classes in the five studies which reported outcomes (random-effect RR, 1.03; 95% CI, 0.83-1.27; p=0.517; I2=0%). Furthermore, there were no significant differences in major adverse cardiac events (MACE), coronary heart disease, myocardial infarction, ischemic stroke and heart failure. However, there were less hypoglycemic events and weight gain in the DPP4 inhibitor group as compared with the SU group (random-effect RR, 3.79; 95% CI, 1.53-9.39; p<0.001; I2=98.2 and weighted mean difference, 1.68; 95% CI, 1.07-2.29; p<0.001; I2=94.7, respectively).
CONCLUSION: As add-on therapy to metformin, there were no significant differences in all-cause mortality and cardiovascular mortality between DPP4 inhibitors and SUs.

INTRODUCTION

Guidelines recommend metformin and comprehensive life style modification as first-line therapy in the treatment of type 2 diabetes mellitus (T2DM).1234 This recommendation is based on the benefits of metformin compared to other class of oral hypoglycemic agents such as sulfonylureas (SUs), thiazolidinediones, and dipeptidyl peptidase-4 (DPP4) inhibitors, with regard to the combined effect on hemoglobin A1c (HbA1c), weight gain, hypoglycemic side effects, socioeconomic burden, and long-term cardiovascular disease.5

In cases where the maximal dose of a single agent cannot maintain HbA1c below 6.5%, combination therapy is recommended.678 Although the recommended specific agent differs according to baseline risk stratification, DPP4 inhibitors, SUs, thiazolidinediones, α-glucosidase inhibitors, and sodium-glucose cotransport-2 (SGLT-2) inhibitors can be used as a combination therapy with metformin.9 When there is a compelling need to minimize hypoglycemia, DPP4 inhibitor can be considered. For patients whose major issue is cost, SUs or thiazolidinediones could be a good choice. For patients with atherosclerotic cardiovascular disease, heart failure or chronic kidney disease, treatment with glucagon-like peptide-1 (GLP1) receptor agonists or SGLT2 inhibitors as combination therapy should be preferred according to individual status. In Korea, DPP4 inhibitor and SU are the most commonly used drugs for add-on therapy with metformin.

DPP4 inhibitor acts by decreasing degradation of GLP1 and glucose-dependent insulinotropic peptide.10 It has been shown to effectively control blood glucose levels and lower level of HbA1C with minimal risk of hypoglycemia and good tolerability.1112 However, previous studies have shown concerns of increased risk of heart failure.1314 On the other hand, SU, a secretagogue of insulin, can effectively decrease blood glucose with neutral effects on the cardiovascular system, but hypoglycemia and weight gain are common side effects.1516 In this current study, we conducted a systematic review through meta-analysis to compare the cardiovascular risk of DPP4 inhibitors with that of SUs during combination therapy with metformin.

MATERIALS AND METHODS

1. Data sources and searches

We performed systemic electronic search of MEDLINE, Cochrane Library, and Embase, with no language limits, using search terms as following: “Diabetes mellitus,” “Sulfonylurea,” “glimepiride,” “Glipizide,” “gliclazide,” “glibenclamide,” “glyburide,” “gliguidone,” “Dipeptydil peptidase-4 inhibitor,” “DPP4 inhibitor,” “Sitagliptin,” “Vildagliptin,” “Linagliptin,” “Saxagliptin,” “Alogliptin,” and “Dutogliptin.”

2. Study selection

Two reviewers (W.K. Jeon and J. Kang) independently searched the articles with the following inclusion criteria: 1) Both randomized control trials (RCTs) and non-randomized trials examining efficacy of combination therapy of metformin with DPP4 inhibitor compared to SU were searched, except case-control studies; 2) The study should analyze cardiovascular risks of combination therapy, including cardiovascular death, myocardial ischemia, and heart failure; and 3) Each study should propose incidence rate of mortality or morbidities. Disagreed articles were resolved by discussion (Supplementary Table 1).

3. Data extraction and quality assessment

Two reviewers independently extracted data from the included articles using a standardized form. By comparing data from each reviewer, internal consistency was examined and inconsistent data was corrected by discussion. Extracted data was all-cause mortality, cardiovascular death, ischemic stroke or transient ischemic attack, serious cardiovascular or cerebrovascular adverse event reported in the study which was regarded as major adverse cardiac events (MACE), coronary heart disease, myocardial infarction, heart failure, hypoglycemic event and weight change.

In case of RCTs, quality was assessed by using the Cochrane Collaboration’s tool for assessing the risk of bias 2.0 for RCTs. Random sequence generation, allocation concealment, blinding of participants, blinded outcome assessment, complete follow-up, and selective reporting were assessed. In case of non-randomized trials, the Strengthening The Reporting of Observational studies in Epidemiology (STROBE) checklist was used.

4. Data synthesis and analysis

Meta-analysis was performed based on the random-effect model. To qualitatively assess for small study bias, funnel plots were constructed. To quantitatively assess, Egger's linear regression method was used. If small study bias was found, the trim and fill method was used. Cochran's Q via a χ2 test and I2 statistics were used to evaluate statistical heterogeneity. The p-values were two-tailed and statistical significance was considered when p<0.05. STATA/SE 12.0 (Stata Corp LP, College Station, TX, USA) were used in statistical analysis.

RESULTS

1. Identification and selection of studies

Our first search yielded 4753 studies from PubMed, Cochrane Library, and Embase (641, 388, and 3,724 studies respectively). After removal of duplicated, irrelevant, or retrospective studies, and short follow-up duration under 1 year, 15 studies were included in our current analysis. Among the studies, one was a prospective observational study and 14 were RCTs. The study selection process is summarized in Fig. 1.

Fig. 1

Diagram for study selection.

2. Description of included trials

The characteristics of included studies are summarized in Table 1. Among the RCTs, 6 studies exclusively used glipizide as the SU, 6 used glimepiride, and 2 used gliclazide. For DPP4 inhibitors, 4 studies exclusively used vildagliptin, 3 used sitagliptin or linagliptin, and 2 used alogliptin or saxagliptin, respectively. The only non-randomized study permitted various SUs and DPP4 inhibitors. Duration of patient follow-up ranged from 52 weeks to over 312 weeks. All patients were diagnosed as T2DM and seven studies required a minimum dose of metformin to be 1,500 mg before considering combination therapy.

Table 1

Characteristics of included studies

Study	Type	SU dose	DPP4 inhibitor dose	Duration	Inclusion criteria	Primary outcome	Quality of evidence (GRADE)
Gitt et al.17	Prospective	Various	Various	52 weeks	Aged ≥40 years	Change in HbA1c from baseline	Low
					T2DM
					Metformin monotherapy
Arjona Ferreira et al.18	RCT	Glipizide	Sitagliptin	54 weeks	Aged ≥30 years	Change in HbA1c from baseline	Moderate
		10 mg bid	25 mg qd		T2DM
					ESRD with dialysis
					HbA1c 7%–9%
Del Prato et al.19	RCT	Glipizide	Alogliptin	104 weeks	Aged 18–80 years	Change in HbA1c from baseline	High
		20 mg qd	12.5–25 mg qd		T2DM
					HbA1c 7–9%
					Metformin ≥1,500 mg
Ferrannini et al.20	RCT	Glimepiride	Vildagliptin	52 weeks	Aged 18–73 years	Change in HbA1c from baseline	High
		6 mg qd	50 mg bid		T2DM
					HbA1c 6.5%–8.5%
					Metformin ≥1,500 mg
Filozof et al.21	RCT	Gliclazide	Vildagliptin	52 weeks	Aged 18–78 years	Change in HbA1c from baseline	High
		320 mg qd	50 mg bid		T2DM
					HbA1c 7.5%–11.0%
					Metformin ≥1,500 mg
Foley et al.22	RCT	Gliclazide	Vildagliptin	104 weeks	Aged ≥18 years T2DM	Change in HbA1c from baseline	High
		320 mg qd	50 mg bid		HbA1c 7.5–11.0%
					Drug naïve
Baptist Galhwitz et al.23	RCT	Glimepiride	Linagliptin	104 weeks	Aged 18–80 years	Change in HbA1c from baseline	High
		4 mg qd	5 mg qd		T2DM
					HbA1c 6.5%–10.0%
					Metformin ≥1,500 mg
Burkhard Goke et al.24	RCT	Glipizide	Saxagliptin	52 weeks	Aged ≥18 years	Change in HbA1c from baseline	High
		20 mg qd	5 mg qd		T2DM
					HbA1c 6.5%–10%
					Metformin ≥1,500 mg
Markku Laakso et al.25	RCT	Glimepiride	Linagliptin	52 weeks	Aged ≥18 years	Change in HbA1c from baseline	High
		4 mg qd	5 mg qd		T2DM
					HbA1c 7.0%–10.0%
					CKD
Matthews et al.26	RCT	Glimepiride	Vildagliptin	104 weeks	Aged 18–73 years	Change in HbA1c from baseline	High
		6 mg qd	50 mg bid		T2DM
					HbA1c 6.5%–8.5%
					Metformin ≥1,500 mg
Rosenstock et al.27	RCT	Glipizide	Alogliptin	52 weeks	Aged 65–90 years	Change in HbA1c from baseline	High
		10 mg qd	25 mg qd		T2DM
					HbA1c 6.5%–9.0% without medication
					or HbA1c 6.5%–8.0% with monotherapy
Schernthaner et al.28	RCT	Glimepiride	Saxagliptin	52 weeks	Aged ≥65 years	Change in HbA1c from baseline	High
		6 mg qd	5 mg qd		T2DM
					HbA1c 7.0%–9.0%
					Metformin any dose
Seck et al.29	RCT	Glipizide	Sitagliptin	104 weeks	Aged 18–78 years	Change in HbA1c from baseline	High
		20 mg qd	100 mg qd		T2DM
					HbA1c 6.5%–10.0%
					Metformin ≥1,500 mg
Arjona Ferreira et al.30	RCT	Glipizide	Sitagliptin	54 weeks	Aged ≥30 years	Change in HbA1c from baseline	Moderate
		2.5–20 mg	25–50 mg qd		T2DM
					CKD
					HbA1c 6.5%–9.0%
Rosenstock et al.31	RCT	Glimepiride	Linagliptin	Median	Adults	CV death Nonfatal MI and stroke	High
		1–4 mg	5 mg qd	6.2 years	T2DM
					HbA1c 6.5%–8.5%
					High CV risk

SU, sulfonylurea; DPP4, dipeptidyl peptidase-4; T2DM, type 2 diabetes mellitus; RCT, randomized controlled trial; bid, twice a day; qd, once a day; ESRD, end-stage renal disease; HbA1c, hemoglobin A1C; CKD, chronic kidney disease; CV, cardiovascular; MI, myocardial infarction.

The result of Cochrane collaboration's risk assessment is shown in Table 2. One prospective cohort study17 was assessed by the STROBE checklist and scored 19.

Table 2

Cochrane risk of bias tools 2.0

Study	1	2	3	4	5	Overall risk of bias
Arjona Ferreira et al.18	Low	Low	Low	Low	Low	Low
Del Prato et al.19	Low	Low	Some concerns	Low	Low	Some concerns
Ferrannini et al.20	Low	Low	Low	Low	Low	Low
Filozof et al.21	Low	Low	Low	Low	Low	Low
Foley et al.22	Low	Low	Low	Low	Low	Low
Baptist Galhwitz et al.23	Low	Low	Low	Low	Low	Low
Burkhard Goke et al.24	Low	Low	Some concerns	Low	Low	Some concerns
Markku Laakso et al.25	Low	Low	Low	Low	Low	Low
Matthews et al.26	Low	Low	Low	Low	Low	Low
Rosenstock et al.27	Low	Low	Low	Low	Low	Low
Schernthaner et al.28	Low	Low	Low	Low	Low	Low
Seck et al.29	Low	Low	Some concerns	Low	Low	Some concerns
Arjona Ferreira et al.30	Low	Low	Low	Low	Low	Low
Rosenstock et al.31	Low	Low	Low	Low	Low	Low

1, bias due to randomization process; 2, bias due to deviations from the intended interventions; 3, bias due to missing outcome data; 4, bias in measurement of the outcome; 5, bias in selection of the reported result.

3. Analysis endpoint outcomes

The primary analysis endpoint of the meta-analysis was all-cause mortality. All-cause mortality data was achieved from 13 studies while Rosenstock et al.31 reported no mortality events during follow-up and thus was excluded in the analysis. There was no significant difference in all-cause mortality among 12 studies comparing SU and DPP4 inhibitor as an add-on therapy to metformin (random-effect relative risk [RR], 1.14; 95% confidence interval [CI], 0.98–1.33; p=0.811; I2=0%) (Fig. 2). Cardiovascular death also showed no significant difference in an analysis of 5 studies which reported cardiovascular death as one of the outcomes (random-effect RR, 1.03; 95% CI, 0.83–1.27; p=0.517; I2=0%) (Fig. 2).

Fig. 2

Meta-analyses for all-cause mortality and cardiovascular mortality.

SU, sulfonylurea; DPP4, dipeptidyl peptidase-4; RR, relative risk; CI, confidence interval; df, degrees of freedom.

Regarding morbidity events, DPP4 inhibitors were associated with a higher risk of ischemic stroke or transient ischemic attack from an analysis of six studies which reported such outcomes (random-effect RR, 2.78; 95% CI, 1.06–7.30; p=0.065; I2=51.9%) (Fig. 3). However, there was no significant difference in MACE (random-effect RR, 1.05; 95% CI, 0.91–1.23; p=0.568; I2=0%), coronary heart disease (random-effect RR, 0.96; 95% CI, 0.79–1.17; p=0.742; I2=0%), myocardial infarction (random-effect RR, 1.04; 95% CI, 0.83–1.29; p=0.640; I2=0%) and heart failure (random-effect RR, 0.90; 95% CI, 0.73–1.12; p=0.839; I2=0%) in analysis of studies which reported outcomes (Fig. 3).

Fig. 3

Meta-analyses for various morbidities; ischemic stroke, MACE, coronary heart disease, myocardial infarction, and heart failure.

MACE, major adverse cardiac events; SU, sulfonylurea; DPP4, dipeptidyl peptidase-4; RR, relative risk; CI, confidence interval; df, degrees of freedom.

Regarding the representative side effects of specific agents, DPP4 inhibitors showed significantly lower risk of hypoglycemic events in analysis of twelve studies which reported such outcomes (random-effect RR, 3.79; 95% CI, 1.53–9.39; p<0.001; I2=98.2%) (Fig. 4) and weight gain in analysis of eight studies which reported such outcomes (weight mean difference, 1.68; 95% CI, 1.07–2.29; p<0.001; I2=94.7%) (Fig. 4).

Fig. 4

Meta-analyses for hypoglycemia and weight gain.

SU, sulfonylurea; DPP4, dipeptidyl peptidase-4; RR, relative risk; CI, confidence interval; df, degrees of freedom; SD, standard deviation; WMD, weighted mean difference.

4. Analysis of small study bias

The presence of small study bias was assessed and confirmed by funnel plot asymmetry and Egger's linear regression method. The funnel plot test of primary analysis end point was presented in Supplementary Fig. 1. The only found small study bias in the analysis was ischemic stroke (bias coefficient=1.387; SE=0.444; t=3.12; p=0.036; 95% CI, 0.153–2.621). After adjustment through the trim and fill method, there was no significant difference in ischemic stroke between SU and DPP4 inhibitor (random-effect RR, 1.28; 95% CI, 0.50–3.29; p=0.612) (Fig. 3).

DISCUSSION

We conducted systematic review of one prospective study and 14 RCTs including the CAROLINA trial, one of the most recent and largest studies with long-term follow-up.31 All of the studies compared the efficacy and safety of add-on therapy on top of metformin between DPP4 inhibitors and SUs for adults diagnosed with T2DM. In our analysis, we found no significant differences in all-cause mortality and cardiovascular mortality between DPP4 inhibitor and SU. Morbidities like MACE, coronary heart disease, myocardial infarction, ischemic stroke and heart failure also showed no significant differences. There was a slight difference in ischemic stroke, but after correction for small study bias, the difference was not significant. There were some endpoints that did show differences between the two drug classes. One was the rate of hypoglycemic event, which was significantly lower with DPP4 inhibitors, and the other was weight gain, which was significantly lower with DPP4 inhibitors. Such favorable results regarding hypoglycemia and weight gain are consistent with previous reports.32 Collectively, our data showed no differences between DPP4 inhibitors and SUs regarding hard clinical endpoints, but regarding side effects, DPP4 inhibitors showed beneficial effect, suggesting the safety and good feasibility of DPP4 inhibitors when add-on therapy is needed on top of metformin.

Several previous placebo-controlled randomized studies have been reported using DPP4 inhibitors as add-on therapy to metformin.143334353637 Although the risk of hospitalization for heart failure was increased in the SAVOR-TIMI 53 trial,14 most studies showed non-inferior cardiovascular outcomes of DPP4 inhibitors compared to placebo. There was no difference in hospitalization for heart failure in the CARMELINA34 and TECOS trials.37

Some researchers had concern about the harmful effect of SU on cardiovascular risk and mortality.3839 Because this concern was not supported always,40 this controversial finding had been debated for a long time.

The results of head-to-head comparisons with DPP4 inhibitor and SU in previous meta-analyses have not been consistent. Some studies showed lower cardiovascular mortality with DPP4 inhibitors,3841 while others showed no significant differences.4243 DPP4 inhibitors showed favorable trend regarding myocardial infarction in two previous meta-analyses,3844 but others reported negative findings.4243 Regarding MACE, 2 studies showed lower risk with DPP4 inhibitors,3242 while one other study showed no significant difference.45 These inconsistent results could be explained by different character of included studies. There is more possibility that studies which are including observational studies have potential bias, than the studies which analyzed only randomized trials. Some studies analyzed in previous meta-analysis showed too much wide CI, which lower reliability of the results. Moreover, most of the studies included in previous meta-analyses were designed to analyze the effect of treatment, not mortality or MACE.

There are several limitations in this study. First, our meta-analysis was not a patient-level but rather a study-level meta-analysis, and thus we did not have individual patient data. Therefore, although we identified 15 studies, not all studies could be included in each analysis. Certain studies reported different combinations of outcomes that we analyzed in the present study. Second, there were significant differences in the weight of the studies included. For example, our analysis heavily depended on the CAROLINA trial because it was the largest trial with the longest follow-up study among the 15 studies. However, we performed a sensitivity analysis excluding the CAROLINA study, and found consistent results across all studies outcomes except for ischemic stroke (Supplementary Figs. 2-4). Third, Most of RCTs were sponsored or funded by pharmaceutical company. Fourth, we did not distinguish each individual drugs within the same class or the different drug doses. Therefore, we could not differentiate whether our findings are a class effect or an effect of a certain individual drug. Efficacy and safety can be different in each individual drugs of same class or by doses of same drugs. Although linagliptin, alogliptin, and sitagliptin showed no significant difference in cardiovascular event or hospitalization for heart failure,33343536 saxagliptin showed increased hospitalization for heart failure.14 Finally, we did not consider other medications which might affect patient's cardiovascular status, like antihypertensive drugs or statins.

In conclusion, there were no significant differences in major cardiovascular outcomes between DPP4 inhibitors and SUs when used on top of metformin. There were slightly beneficial effects of DPP4 inhibitors such as lower rates of hypoglycemia and less weight gain, suggesting good safety and feasibility of the drugs. DPP4 inhibitors can be a good option as add-on therapy to metformin in patients with T2DM. Also, regarding cardiovascular risk and cost-effectiveness, SU can be a reasonable alternative.